Question

Example tables for SQL Server.

• MainRecords (id, record)
• AuxRecords (mainRecords_id, record)
• SourceRecords (record1, record2)

MainRecords.id is a self incrementing primary identity key.

Is it possible to select from SourceRecords and insert into MainRecords and AuxRecords at the same time such that MainRecords.record = record1, AuxRecords.record = record2, and AuxRecords.mainRecords_id = MainRecords.id all in a single statement?

EDIT:

Based on tip from below, I tried this…

DECLARE @MainRecords table(id int PRIMARY KEY IDENTITY, record varchar(5))
DECLARE @AuxRecords table(mainRecords_id int, record varchar(5))
DECLARE @SourceRecords table(record1 varchar(5), record2 varchar(5))

INSERT @SourceRecords VALUES ('a', 'a')
INSERT @SourceRecords VALUES ('a', 'b')
INSERT @SourceRecords VALUES ('a', 'c')
INSERT @SourceRecords VALUES ('b', 'a')
INSERT @SourceRecords VALUES ('b', 'b')
INSERT @SourceRecords VALUES ('b', 'c')
INSERT @SourceRecords VALUES ('c', 'a')
INSERT @SourceRecords VALUES ('c', 'b')
INSERT @SourceRecords VALUES ('c', 'c')

INSERT INTO @MainRecords (record)
  OUTPUT inserted.id, @SourceRecords.record2
  INTO @AuxRecords (mainRecords_id, record)
SELECT record1 FROM @SourceRecords

select * from @MainRecords
select * from @AuxRecords

But unfortunately get error:

Msg 137, Level 16, State 1, Line 16
Must declare the scalar variable "@SourceRecords".

If I change those table type variables into real tables I get the error:

Msg 4104, Level 16, State 1, Line 3
The multi-part identifier "SourceRecords.record2" could not be bound.

Below works fine, but obviously it isn’t a complete solution. I’m just showing it to demonstrate that my syntax is correct.

INSERT INTO @MainRecords (record)
  OUTPUT inserted.id --, @SourceRecords.record2
  INTO @AuxRecords (mainRecords_id) --, record)
SELECT record1 FROM @SourceRecords

So… unless there is some trick I am missing, it seems like OUTPUT is a dead end solution for this problem.

The other suggestions of creating views or empty tables with triggers are not a “single statement” solution to the problem. Moreover, they add obscurity, whereas adding some extra columns and using a stored procedure is equally complex but more obvious and straightforward.

Answer 1

community wiki

there isn’t enough info in the question to write code to solve the question. So, here is a “generic” output clause example, which does not relate at all to this question, other than to show how OUTPUT can be used:

this will delete, insert, and return multiple rows in a single statement

DECLARE @OldTable table(col1 int, col2    varchar(5), col3 char(5), col4     datetime)
DECLARE @NewTable table(col1 int, column2 varchar(5), col3 int    , col_date char(23), extravalue int, othervalue varchar(5))
INSERT @OldTable VALUES (1 , 'AAA' ,'A'  ,'1/1/2010'           )
INSERT @OldTable VALUES (2 , 'BBB' ,'12' ,'2010-02-02 10:11:22')
INSERT @OldTable VALUES (3 , 'CCC' ,null ,null                 )
INSERT @OldTable VALUES (4 , 'B'   ,'bb' ,'2010-03-02'         )

DELETE @OldTable           --<<<alter table 1
    OUTPUT DELETED.col1    --<<<alter table 2
          ,DELETED.col2
          ,CASE
               WHEN ISNUMERIC(DELETED.col3)=1 THEN DELETED.col3 
               ELSE NULL END
          ,DELETED.col4
          ,CONVERT(varchar(5),DELETED.col1)+'!!'
        INTO @NewTable (col1, column2, col3, col_date, othervalue)
    OUTPUT 'Rows Deleted: ', DELETED.* --<<<returns a result set
    WHERE col1 IN (2,4)

SELECT * FROM @NewTable

OUTPUT:

               col1        col2  col3  col4
-------------- ----------- ----- ----- -----------------------
Rows Deleted:  2           BBB   12    2010-02-02 10:11:22.000
Rows Deleted:  4           B     bb    2010-03-02 00:00:00.000

(2 row(s) affected)

col1        column2 col3        col_date                extravalue  othervalue
----------- ------- ----------- ----------------------- ----------- ----------
2           BBB     12          Feb  2 2010 10:11AM     NULL        2!!
4           B       NULL        Mar  2 2010 12:00AM     NULL        4!!

(2 row(s) affected)